feat: integrate lifecycle event broadcasting into DevicePollerWorker

Add Events.broadcast_job_event calls to track job lifecycle:
- :started when job begins executing
- :completed/:failed when job finishes
- Track duration in seconds for all job executions

Broadcast events enable real-time monitoring of polling operations.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

lib/towerops/workers/device_poller_worker.ex

@@ -22,6 +22,7 @@ defmodule Towerops.Workers.DevicePollerWorker do
 
   alias Towerops.Agents
   alias Towerops.Devices
+  alias Towerops.JobMonitoring.Events
   alias Towerops.Snmp
   alias Towerops.Snmp.ArpDiscovery
   alias Towerops.Snmp.Client
@@ -34,17 +35,33 @@ defmodule Towerops.Workers.DevicePollerWorker do
   @default_poll_interval 60
 
   @impl Oban.Worker
-  def perform(%Oban.Job{args: %{"device_id" => device_id}}) do
-    case Devices.get_device(device_id) do
-      nil ->
-        Logger.debug("Device #{device_id} no longer exists, skipping poll")
-        :ok
+  def perform(%Oban.Job{args: %{"device_id" => device_id}} = job) do
+    Events.broadcast_job_event(job, :started)
+    start_time = System.monotonic_time(:second)
 
-      device ->
-        maybe_poll_device(device)
-        schedule_next_poll_with_error_handling(device_id, device)
-        :ok
+    result =
+      case Devices.get_device(device_id) do
+        nil ->
+          Logger.debug("Device #{device_id} no longer exists, skipping poll")
+          :ok
+
+        device ->
+          maybe_poll_device(device)
+          schedule_next_poll_with_error_handling(device_id, device)
+          :ok
+      end
+
+    duration = System.monotonic_time(:second) - start_time
+
+    case result do
+      :ok ->
+        Events.broadcast_job_event(job, :completed, %{duration: duration})
+
+      {:error, reason} ->
+        Events.broadcast_job_event(job, :failed, %{error: inspect(reason), duration: duration})
     end
+
+    result
   end
 
   defp maybe_poll_device(device) do